《计算机应用研究》|Application Research of Computers

基于l2,1范数原子选择的图像分块稀疏重构

Image segmentation sparse reconstruction based on l2,1-norm atomic selection

免费全文下载 (已被下载 次)  
获取PDF全文
作者 朱华,岳峻,李振波,张志旺,寇光杰
机构 1.鲁东大学 信息与电气工程学院,山东 烟台 264025;2.中国农业大学 信息与电气工程学院,北京 100083
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2019)05-058-1560-04
DOI 10.19734/j.issn.1001-3695.2017.12.0829
摘要 针对图像压缩采样中原子的选择规则难以确定的问题,在改进的正交匹配追踪算法的基础上提出了一种基于l2,1范数的原子选择方式。l2,1范数的原子选择方式考虑了原子间的相关性,剔除了干扰原子,选择出了代表性原子。将所提方法用于图像分块重构,算法以图像进行分块,利用l2,1范数选择对图像块支撑集进行筛选,增强块特征的判别性,提高原子的稀疏度,最终提高图像重构的准确率和速率。实验结果表明,相同条件下在保证重建速度的同时,所提新方法提高了图像重构精度。
关键词 压缩感知; 稀疏表示; l2, 1范数选择; 图像重构; 图像分块; 匹配追踪
基金项目 国家自然科学基金资助项目(61472172,61673200,61471185)
山东省自然科学基金资助项目(ZR2017MF062,ZR2016FM15)
本文URL http://www.arocmag.com/article/01-2019-05-058.html
英文标题 Image segmentation sparse reconstruction based on l2,1-norm atomic selection
作者英文名 Zhu Hua, Yue Jun, Li Zhenbo, Zhang Zhiwang, Kou Guangjie
机构英文名 1.College of Information & Electrical Engineering,Ludong University,Yantai Shandong 264025,China;2.College of Information & Electrical Engineering,China Agricultural University,Beijing 100083,China
英文摘要 In order to solve the problem of selecting atoms for image compressed samples automatically, this paper proposed a new l2, 1-norm atomic selection method. The proposed method took into account the correlation between the atoms, eliminated the interference of atoms and selected a representative of the atomic. Image reconstruction based on image segmentation and l2, 1-norm selection. This algorithm divided image into blocks, and filtered the image blocks supporting set by the l2, 1-norm selection which enhanced the discriminant of block features, improved the sparsity of atoms, and ultimately improved the accuracy and speed of image reconstruction. The simulation results show that the reconstruction accuracy can be improved with the same condition.
英文关键词 compressed sensing; sparse representation; l2, 1-norm selection; image reconstruction; image block; matching pursuit
参考文献 查看稿件参考文献
 
收稿日期 2017/12/4
修回日期 2018/1/12
页码 1560-1563
中图分类号 TP391.41
文献标志码 A